Extraction and analysis of digital images feature of three kinds of wheat diseases

In this paper, the method of automatic identification of three wheat diseases was applied by analyzing the morphological characteristics extracted from their images. The target area was got by segmenting three kinds of wheat diseases images on wheat powdery mildew, wheat sharp eyespot, and wheat stripe rust. Via extracting and optimizing the morphological data and using statistical analysis software to analysis the data with the principal component analysis and the discriminant analysis, five characteristic parameters such as Sphericity, Roundness, Hu1, Hu2, equivalent radius were selected as the identification factors. The recognizable rates of the samples among the three wheat diseases were 96.7%, 93.3%, and 86.7% respectively using the factors.